The Japanese justice system is undergoing wide-ranging reform. The Office for Promotion of Justice System Reform working within the Cabinet has just today expressed an “urgent need for an efficient and systematic mechanism to translate Japanese legislation into foreign languages in response to calls from lawyers and businesspeople.” The communiqué goes on:

Despite Japan being the world’s second-largest economy, the government has no unified system (sic) to translate Japanese laws into English and other languages. Some major laws have been translated by related ministries or the private sector, but the quality of the translation is often poor and terminology varies.

The group under the Office for Promotion of Justice System Reform of the Prime Minister’s Office suggests that government-led translation of the laws is necessary to facilitate international trade and promote foreign investment in Japan, as well as to assist foreign residents in the country.

“It has long been a big problem that Japanese laws have not been translated coherently and systematically,” said Noboru Kashiwagi, a professor of law at Tokyo’s Chuo University and a member of the group. “Today marks the first step and it is very meaningful.”

The proposal also advises utilizing a computerized system developed by Nagoya University, which includes an electronic dictionary and a database of past translations. But the group has yet to reveal an estimated cost for the project.

It remains undecided whether the task should be taken up solely by the government, to be commissioned to the private sector, or to be done by establishing a new quasi-governmental organization. But the group is envisioning a public-private collaboration.

Now, would it be worth while suggesting to Japan that they check on how the European Union has handled a similar problem – the translation of Community law into local Member State languages as part of membership commitment? Probably not. As far as I know, there is no ’systematic mechanism’ or ‘unified system’ in place to help EU translators through the immense task of localizing the canonical 85,000 pages of EU law. In the past, though, this very process has largely contributed to the development of a translation ‘culture’ in countries such as Denmark, Sweden, Finland and Greece, and has usually been handled via government contracts to legal translators, working in groups.

Since this translation work is ruled by the subsidiarity principle, it is up to each country (rather than a central agency) to get the job done. The exceptional nature of the EU situation lies precisely in having a Member State translate and integrate EU legislation into its own national law and language. Almost by definition, nations normally draft their own laws, rather than localize someone else’s. And here lies the rub: the process involves translators, lawyers, revisers, and anyone else working to ensure that the resulting texts are legally watertight, and optimally adapted to the local context. A hard process to automate as a unified system, but worth attempting surely. EU legal translation is, if you like, inward bound, exclusively for the use of the new Member State.

In the case of Japan, however, it looks as if the translation is outward bound to the world at large. Yet presumably the proposed ‘English’ translation of Japanese law has to use English terms from somewhere – the U.S., Canada, the U.K., Australia etc, and is therefore implicitly localizing to a legal culture, if not a specific country. For an example of the translation of the Arbitration Law, see here.

The patent lack of systematic mechanisms or unified systems for localizing EU law was brought home to me the other day by a report in Eurolang, the news site for Lesser Used Languages in Europe, which proudly stated that Slovenia, one of the newest members of the EU, has developed some ‘computer-based translation tools’ to help in its EU translation effort. What Slovenia has actually done is to develop “a collection of 62,000 technical-term equivalents, called Evroterm (using Trados’ Multiterm), and a bilingual (English-Slovenian) aligned corpus of translations of legal acts by the EU, called Evrokorpus (currently 8.7 million words). You can access them at (SVEZ) website. But I would call these ‘digital resources’ rather than tools.

What is astonishing to me is that this should be news. Why is it that after a decade of European Commission language technology R&D programs, some with multi-million euro funding, we have no better technology to handle these fundamental translation tasks than a term management system first built in the late 1980s and standard corpus (or translation memory) software? I’m sure that the experience of other Member State translation efforts has propagated some form of best practice for localizing Lex europae. I am equally sure that the reason why that there is no rich corpus of bilingual translations in the newer EU countries to feed translation memories and help automate some of the slog, is because national legal content has typically not needed to be translated.

But looking back, it seems a singular error for EC technology czars to have failed to fund innovative projects (even if they proved inadequate) to this critical EU task. They might just have come up with a more “unified system” (e.g. a set of legal ontologies to drive a translation engine) to help new Members get this job done quicker, better, cheaper. And they might have been able to promote it to the Japanese to help with their legal translation projects. Come to think of it, the Japanese too sunk trillions of yen in 5th Generation computing, producing electronic dictionaries, machine translation systems, knowledge bases and so on. All to no apparent avail. Lex dura indeed, sed lex – and highly lexical with it.

Alessandro Barrico, the Italian translator of Homer, has outperformed The Da Vinci Code in sales in Italy with his new Italian version of the Iliad, published by Feltrinelli.

Enthusiasm for the book has led to public readings, starting 2,600 people listening for 12 hours a night for three nights to the spoken translation in Rome. On October 8, the book’s 20,000 or so lines were read in relays by people from different backgrounds in Verona for 24 hours. A new. The next step in this mass vote for the old bard and his martial themes could be to memorize the whole thing.

It wouldn’t be the first time some one has tried, though. Although it’s hard to find out much about it, a retired U.S. businessman called Steven V.N. Powelson decided to memorize the whole Iliad, and was able to perform vast chunks of it around 1994. This meant getting your personal Mnemosyne to get its aural archive around some 200,000 spoken syllables. Powelson took 16 years, working about an hour a day (making a total of 5,840 hours) as befits a 76 year old. He apparently used to recite it as part of a national tour to promote the classics, but I cannot discover whether he ever stood up like Italy’s war-scarred public (Iraq not Troy) and performed the whole thing from beginning to end. For a weird commentary on the ‘autoerotics’ of memorizing poems see this

Here’s how Barrico summarized his motivations:

Quando racconto questo lavoro, spesso la gente mi chiede: perché proprio l’Iliade? Alcuni vorrebbero l’Odissea (che io non amo, tranne il finale), o magari Dante o Ariosto. Ho due risposte: la prima è che l’Iliade mi sembra una storia bellissima. La seconda è che godere del racconto di una guerra mi sembra una cura efficace per allontanare il desiderio (tragico ma legittimo) di godere facendo la guerra.

Check this IDG article on OpenOffice.org localization for three South African languages – Zulu, Afrikaans and Sepedi, plus information on Microsoft’s Kiswahili Local Language Program. The MS initiative also draws on local community contributions to produce a glossary of around 3,000 IT terms in the language.